The Advancement of X-Ray Screening Technologies

The increasing threat of terrorist attacks in recent years has led to large investments in technological advancements in aviation security. One main focus continues to be the improvement of the process of X-ray screening of passenger bags to prevent prohibited items to get past security checkpoints (Brady). Some airports have started using the newest technologies of cabin baggage screening, such as multi-view X-ray systems. Some of these new systems provide automated detection of explosives, leading to a substantial improvement in security. However, human operators are still needed to visually inspect X-ray images (Bülthoff, Edelman).

The task of threat detection in objects depends on knowledge-based and image-based factors. Knowledge-based factors refer to knowing which items are prohibited and what they look in an X-ray image. Some objects like the stun gun are difficult to differentiate from commonly used electronics like a cell phone or a hard drive when shown on an X-ray image (Cohen).

Image-based factors also play an important role in threat object detection. This is often attributed to the visual abilities of a person, that is, how he or she handles image difficulty. There are three main image-based factors: rotation, superposition, and bag transparency (Graf, Schwaninger, Wallraven). Rotation refers to when forbidden objects are difficult to recognize from an unusual viewpoint and distinctive features are not visible. Another important factor is the superposition of the threat by other objects in a bag. If a threat object, such as a knife, is superimposed by a high density material, it becomes difficult to recognize the distinctive shape of the object. The transparency of a bag, determined by the number and type of objects in the bag, has a significant influence on detection performance making it difficult to distinguish prohibited items (Green, Swets).

Besides issues with knowledge-based and image-based factors, the screener also has to detect a prohibited item in a limited amount of time. During peak hours at a busy international airport, screeners often have only a few seconds to visually inspect the X-ray images before moving on to the next bag. There is also evidence that perceptual training can help screeners distinguish prohibited objects from cluttered visual scenes. Through training, screeners can learn how different prohibited objects appear in X-ray images (Hardmeier, Michel, Schwaninger).

A study was conducted in Zurich airport where 32 different novices participated in 512 trials (single-view and multi-view mixed). Their task was to determine whether the bag presented was OK (didn’t contain a threat item) or if the bag was NOT OK (contained a threat item) by clicking the respective button on the screen. In addition, participants were asked to indicated how confident they were in their decision by clicking on a rating scale on the screen (Kourtzi, Betts, Sarkheil, Welchman).

Overall, the experiment showed that the multi-view X-rays made it easier to detect prohibited items when they were rotated or superimposed with other objects.

Figure 1: X-ray image of a stun gay looks familiar to a cell phone or electronic device.

Figure 2: Image based factors with an impact on threat detection performance
(a) Easy and difficult rotation
(b) Low and high superposition
(c) Low and high bag transparency

References:

Brady, M. J.; Kersten, D.: Bootstrapped learning of novel objects. Journal of Vision, 3(6), pp. 413 – 422, 2003

Bülthoff, H. H.; Edelman, S.: Psychophysical support for a 2-D view interpolation theory of object recognition. Proceedings of the National Academy of Sciences of the United States of America, 89, pp. 60 – 64, 1992

Cohen, J.: Statistical Power Analysis for the behavioral sciences (2nd ed.). Lawrence Erlbaum Asso- ciates, Hillsdale, NJ, 1988

Graf, M.; Schwaninger, A.; Wallraven, C.; Bülthoff, H. H.: Psychophysical results from experi- ments on recognition & categorization. Information Society Technologies (IST) programme, Cognitive Vision Systems – CogVis (IST-2000-29375), 2002

Green, D. M.; Swets, J. A.: Signal detection theory and psychophysics. Wiley, New York, 1966

Hardmeier, D.; Hofer, F.; Schwaninger, A.: The X-ray object recognition test (X-ray ORT) – a reli- able and valid instrument for measuring visual abilities needed in X-ray screening. IEEE ICCST Proceedings, 39, pp. 189 – 192, 2005

Ishihara, S.: Ishihara’s tests for colour-blindness. Kanehara and Co, Tokyo, 2003

Koller, S. M.; Hardmeier, D.; Michel, S.; Schwaninger, A.: Investigating training and transfer ef- fects resulting from recurrent CBT of X-ray image interpretation. In D. S. McNamara & J. G. Trafton (Eds.), Proceedings of the 29th Annual Cognitive Science Society, Cognitive Science Society, Austin, TX, 2007

Koller, S. M.; Hardmeier, D.; Michel, S.; Schwaninger, A.: Investigating training, transfer, and viewpoint effects resulting from recurrent CBT of X-ray image interpretation. Journal of Transportation Security, 1(2), pp. 81 – 106, 2008

Kourtzi, Z.; Betts, L. R.; Sarkheil, P.; Welchman, A. E.: Distributed neural plasticity for shape learning in the human visual cortex. PLoS Biology, 3(7), pp. 1317 – 1327, 2005